When it comes to sustainable energy, harnessing nuclear fusion is—for many—a holy grail of sorts. Unlike climate-warming fossil fuels, fusion offers a clean, nearly limitless source of energy by combining light atomic nuclei to form heavier ones, releasing vast amounts of energy in the process.
But it isn’t easy replicating and controlling the process that powers the sun.
“We eventually want to move to producing energy this way,” says Brian Leard ’21, a fifth-year PhD student in Lehigh University’s Department of Mechanical Engineering and Mechanics. “But you have to heat the plasma fuel to really, really high temperatures and densities, and then you have to confine it. That’s extremely difficult.”
As part of the legion of researchers around the world attempting to solve this and other problems associated with turning the vision of nuclear fusion into reality, Leard was recently recognized by the U.S. Department of Energy through its prestigious Office of Science Graduate Student Research program (SCGSR). According to the agency, “the program helps prepare doctoral candidates to enter jobs of critical importance to the DOE mission and secure the nation’s status at the forefront of discovery and innovation.” From a large, diverse, and highly competitive field of applicants, DOE selected just 62 students from 24 states to pursue their research at the agency’s national laboratories, and ultimately, contribute to addressing the country’s energy, environmental, and nuclear challenges.
Leard’s research project, “Toward Dual Plasma Equilibrium and Transport Optimization,” will be conducted at the DIII-D National Fusion Facility operated by General Atomics in San Diego.
The project focuses on optimizing the performance of tokamak reactors—devices designed to confine hot plasma (which serves as the fuel for generating nuclear reactions) using magnetic fields. The fusion fuel, typically isotopes of hydrogen like deuterium and tritium, must be heated to temperatures about seven times the temperature at the core of the Sun, necessitating precise control over various parameters to maintain stability and efficiency.
“When you’re trying to reach the conditions necessary to actually start generating energy, there are all sorts of actuators that drive aspects like current, torque, and heating, and change the shape of the plasma as well as its magnetohydrodynamic equilibrium and internal properties,” says Leard. “You have a target state for the plasma, but you don’t know the exact trajectories required by each of the actuators to achieve those conditions.”
Together with his team at Lehigh—which is led by Eugenio Schuster, a professor of mechanical engineering and mechanics in the P.C. Rossin College of Engineering and Applied Science and Leard’s advisor—the PhD student has developed a novel method that integrates advanced neural network models with plasma physics to develop simulation codes that can be used to optimize actuator operation. Such simulation has been done before, says Leard, but the codes currently in use often are very time consuming or neglect certain physical processes that can limit their scope and or accuracy.
“The simulation codes we have now capture a lot of the plasma physics, but they generally operate at timescales of hours to days,” he says. “The faster ones make significant assumptions that can impact the prediction accuracy. I’m trying to create a code that is both fast enough for optimization, and more accurate than those previously used.”
Leard and his team are integrating neural-network based surrogate models that are trained on high-fidelity codes that allow them to capture their prediction accuracy without being burdened by their computation time. They’ve also integrated an equilibrium code with a transport code so they can evolve the plasma shape along with the plasma internal profiles like temperature and density, both of which improves the accuracy of their code and allows them to do dual transport and shape optimization.
In San Diego, Leard will focus on conducting the optimizations with this improved simulation code by applying different optimization techniques and different couplings between the optimizer and the simulator.
Such a model could also potentially help reduce the costs of research conducted at tokamaks like DIII-D (a purely experimental machine meant to support the design and operation of future nuclear fusion production plants, like ITER in France)—and lead to new discoveries.
“Every discharge on a tokamak is hundreds of thousands of dollars,” he says. “If we’re able to come up with solutions—meaning the right combination of inputs to get to the desired state we’re looking for—by using a simulation code and an optimizer instead of experimentally, we could also potentially find improved paths toward advanced operating regimes. The algorithm could come up with solutions that we as humans hadn’t considered.”
The SCGSR award will support Leard’s work at the Southern California facility for approximately nine months. For Leard, it’s a unique opportunity to work with national lab researchers and get a feel for that particular career path. And while his work there will be computational, he’s excited about the proximity he’ll have to the DIII-D tokamak.
“I’ll be able to observe the experiments that others are running on the machine,” he says, “and get a better sense of the steps it would take to apply my own work in that way, because ultimately, running experiments is what I want to do.”
Leard was on an airplane, about to take off for a hard-earned vacation, when he found out he’d received the award. And while he says the win was unexpected, he knew he was applying from a position of strength, having Schuster’s guidance and his team’s support from the very beginning of his graduate experience. It was what he calls a “whole group effort.”
“I believe that nuclear fusion represents the future when it comes to our energy ecosystem, and it’s an honor to see that people are interested in my research,” he says. “It motivates me even more to try to solve some of these challenges.”
